to me ,a valid point for additional input would be to actually not only shake in z direction .theres a difference between sustained lateral and longitudinal gs and vibrations there as well the second one surely a point you could evaluate on a rig.
Also there is a difference for all those geometrical load paths in the suspension
for example anti squat or anti dive influencing the behavaviour only dynamically on the road ,or track variation on bump,how are these accounted for when putting the thing on the shaker ?
I missed this interesting thread until now. Many polarized views have been posted covering several aspects of rig testing. Examples are (roughly translated):
- Drivers are a waste of time.
- Drivers' views are the only things that matter.
- Rig testing is the only way to set up vehicle suspensions.
- Math modelling is the only way to set up vehicle suspensions.
- Track testing is the only way to set up vehicle suspensions.
All, in my view, are dangerously wrong, although each contains an element of truth. Again in my view, they illustrate the compartmentalised organisation of many race teams (particularly in F1).
A small example from around 10 years ago:
I was called to help an F1 team with a problem. I found myself in the middle of a disagreement between two factions. Their lead driver & his race engineer had migrated from an initial rear spring stiffness of around 750 lb/in (131 N/mm) to a value of 2300 lb/in (403 N/mm). A rig test demonstrated that the change was wrong (for mechanical set-up), but the race engineer said the change was to reduce rolling at the rear axle. It turned out that a structural change to the vehicle had introduced rear axle compliance steer, causing the chassis to "sideslip" under lateral loads, which interfered with a primary driver cue. A back-back track test showed that the change was preferred by both drivers, but lap times were affected adversely by around a second a lap.
Conclusion: The driver was correct in identifying the problem (his views mattered), but was incorrect in his interpretation (the cockpit is not the best place to identify the source of a problem). His engineer was wrong in his interpretation of the driver's complaints (though his employment prospects would not have been improved by telling the driver he was wrong). Modelling did not identify the problem (because neither the compliance nor driver sensitivity to rear steer was modelled). Rig tests did not help to identify the issue (although they did lead to the problem being identified by a process of elimination).
Another example:
"Aero" race vehicles around the world usually use front springs that are too stiff "mechanically". One of my rig test customers was open to an alternative strategy, so we dropped the front springs by 30 percent, re-optimized the dampers, had a first cut at installing bump rubbers to control ride height at high speeds & set bars to restore lateral balance at low speeds. The team started the next race weekend on their preferred (stiff) set-up, were nowhere after second practice, threw on the "rig set-up" & (as luck would have it) qualified 3rd & won the race.
Conclusion: Probably no race engineer would have made a 30% reduction in front springs at a track. If he had, he probably wouldn't have changed dampers, nor supported the vehicle correctly at high speeds. An example where rig tests were able to help track performance directly.
Having said that, "hardware in the loop" (aka rig) tests are not able (in general) to optimize track performance, for various reasons. They can, however, improve the understanding of a vehicle, helping designers, math modellers and race engineers if they are prepared to pay attention (I don't think I have ever seen a race engineer, for example, attend an F1 rig test).
It is essential (in my view) to use real hardware when rig testing. It is quite common, for example, to use slave engines for various reasons - e.g. for the attachment of hydraulic D/F actuators in an 8 post set-up. This can mask real issues. In one case I proved (eventually) that the engine dissipated almost 40% of the energy input to a vehicle during a rig test. The reason was a poor suspension setup that prevented the dampers working efficiently & the consequence was poor engine reliability.
To sum up, rig tests, math modelling & track tests (including driver feedback) are all important in optimizing a vehicle set-up, but they are complementary tools, not ends in themselves. Hydraulic D/F actuators are essential for rig-based track simulations, but they should be used with care because they can be the most efficient dampers attached to the vehicle if their force control loops are not set accurately. Rig-based track simulations (in my experience) are not particularly revealing, although they can play a part in setting bump rubbers, packer gaps & checking that dampers don't cavitate. Statistical measures (e.g. RMS load variations) are of limited value because, for example, the single kerb input that can launch a vehicle into a wall pales into insignificance when assessed statistically. I have made no mention here of tyres in deference to Jersey Tom, but I do have them in mind when I state "hardware-in-the-loop" tests should have, er, real hardware in the loop, however stylized the tests.
Great post Dave. I'll declare an interest that I've had two sets of my tyres tested on Dave's rig and modified specs based on the results. I've also seen the same set of tyres generate more grip and grain less by putting a DW damper package on the car.
F1_eng - if you really are an F1 engineer it makes me pretty sad that you're so dismissive of drivers. In the end they are the ones who keep the thing on the island and generate the results.
Yes they're bad at feeling high frequency things relating to damping - that's why rig tests are so good. Yes they embelish (I won't say lie) but that's part of the skill of trackside engineering as opposed to driving a workstation - understanding the people you're working with as much as the engineering.
Try telling a MotoGP rider that the computer says a setup's better and you wouldn't last long.
DaveW wrote:Conclusion: The driver was correct in identifying the problem (his views mattered), but was incorrect in his interpretation (the cockpit is not the best place to identify the source of a problem).
I wish more people realized this.
Grip is a four letter word. All opinions are my own and not those of current or previous employers.
Best example for me was putting a really soft compound on that generated good grip but decreased cornering stiffness. We had a correvit and could see the grip level was there but the slip angle was a couple of degrees higher. Driver said "less grip" and he was only wrong if you assume a driver means friction level when he says grip. In this case control moment derivative Calpha x a was probably more relevant to his feeling. The driver is never "wrong" he just might not use the same language as an engineer.
Similarly when bike testing I lost count of the amount of times I'd ask a rider "which tyre felt stiffest" and he'd answer with the construction that was in spring rate terms the softest. In this case decent frictional grip (assessed from descritising the GG diagram) from a soft carcass was being felt as a stiffness because the bike responded quicker.
You have explained it perfectly.
I like to refer to it as the 'fifth element'.
Like in the Bruce Willis film of that name. I should be so lucky!
To run a team, you need enough money, the right people in the right places but without the 'fifth element' it has no hope of coming together.
Many moons ago (before laptops & Pi Research), my group at Cranfield was asked by Team Lotus to instrument a race car. Later, I was privileged to analyse recordings of three drivers in the same vehicle circulating Paul Ricard. All three drove the vehicle completely differently but, surprisingly, returned very similar lap times. Mario Andretti (whose car it was) used all controls like switches, whilst Ronnie Peterson, his team mate, used exaggerated steering inputs to "upset" the vehicle. The real surprise was Jackie Stewart, who had retired from F1 racing a few years earlier. He used very "rounded" inputs, braked earlier, pulled 1.5 gn less around each corner & was (for example) 15 mph slower through Signes (JYS braked, whilst Mario took it flat).
It was an early lesson for me that there is more than one "quick" way around a circuit, and that optimal suspension set-ups are very likely to be different for different driving styles. It also demonstrated that lap times are determined as much by corner exit speeds as they are by braking and corner entry & a driver who fails to deliver an "honest" lap is likely to be denying valuable information both to himself & to his race engineer.
A quote from Chapman about Peterson I could not find translated in English:
"Era extraño. Podías cambiar algo fundamental al coche ¡y él seguía girando en los mismos tiempos! Cuando le preguntaba si notaba algo diferente, respondía: `Pues... derrapa un poco más...´ ¿De dónde? ¿De delante, de atrás, de ambos ejes? `Pues... no estoy seguro...´"
My translation: It was weird. You could change something fundamental on the car ¡and he still lapped in the same times! When I asked if he noticed something different, he answered: "well... it slides a little more..." ¿where? ¿front, rear, both? "well... Im not sure"...
This driver had got such a talent he "covered" the (let me call it) "flaws" of the car changing his inputs, to achieve the (arguably) same final result.
DaveW, thank you very much for sharing the anecdote!
"You need great passion, because everything you do with great pleasure, you do well." -Juan Manuel Fangio
"I have no idols. I admire work, dedication and competence." -Ayrton Senna
Sorry for the earlier confusion. The 8 post rig is just like the video posted. When I read the article I thought they were adding another post to the 7 post rig.
Before I do anything I ask myself “Would an idiot do that?” And if the answer is yes, I do not do that thing. - Dwight Schrute
it is not surprising to me not at all.You can see that all 3 guys were able to get the most out of the car ,at least for one lap.this shows only the potential of the car for that particular time .On a longrun things would likely change towards minimalist inputs and avoiding extended overstressing ,maybe lower peaks ,maybe less time spent under lateral load both could get you more out of the rubber available.
I as well think if petersons car was used for the same test ,with no change to the car allowed possibly only peterson would have been able to extract a top time from it...
best post I have seen in a long time. If this was a ball sport, we would have just picked you up and carried you off the field to the roars of the fans.
Clarity is something we see little of here.
Before I do anything I ask myself “Would an idiot do that?” And if the answer is yes, I do not do that thing. - Dwight Schrute
Giblet wrote:DaveW
best post I have seen in a long time.
agreed, probably the best combination of layman's language, but with clear detail too. up there with post of the year for me.
Would a rig benefit in accuracy from having rotational floors under each tyre? The thought being that turning tyres react differently from stationary ones..
Any post(s) made by this user are (semi-)educated opinion(s), based on random fact(s) blurred by the smudges of time.
Any fact(s) claimed by this user will be supplemented by a link to the original source of said fact(s).
Fil wrote:
Would a rig benefit in accuracy from having rotational floors under each tyre? The thought being that turning tyres react differently from stationary ones..
It probably would be more representative, at least in theory. I am no tyre expert, but I recognise that static tyres "flat spot" when worked on a rig, thus increasing their observed stiffness, & geometries that cause tyres to scrub also lead to increased tyre stiffness estimates. On the other hand, rotating race tyres would probably "flat band" and, if they were rotated at representative speeds, would also self-heat. Under laboratory conditions they might well need to be cooled. I guess that wheel bearings might be happier if they were rotated too. So, several pros & cons. But the main disadvantage of rotating wheels, from my perspective, would be the potential corruption of contact patch load and accelerometer measurements caused by wheel & "roadway" imbalance.
